BANTER for Beaked Whales: EPacific (CCES survey)

###Organizational Instructions Save this file in an R Project that includes the following foldering: /Project |-Database/ |-Binaries/ |-Figures/ |-epacific |_ beakerBanter_epacific.Rmd |_ epacific_banter_data.rds |_ epacific_banter_data_ici.rds |_ epacific_banter_model_ec_t1e4s3_t1e4s4.rds |_ epacific_banter_model_ici_t1e4s3_t1e4s4

PAMpal Data Processing

Start by loading the required packages

library("easypackages")

## Warning: package 'easypackages' was built under R version 4.2.3

libraries("PAMpal", "banter", "rfPermute", "kableExtra", "magick", "magrittr", "here")

## Warning: package 'magick' was built under R version 4.2.3

here()

## [1] "C:/Users/shannon.rankin/Documents/GitHub/BANTER_BeakedWhales"

1. Set up our PPS (PAMPal Settings Object) for the EPacific dataset

pps <- PAMpalSettings(db='Databases/', 
                      binaries = 'Binaries/',
                      sr_hz='auto', 
                      winLen_sec=.0025, 
                      filterfrom_khz=10, 
                      filterto_khz=NULL)

2. Process data and save to file to eliminate repeated processing.

If this is the initial processing, ensure you have set ‘freshRun = TRUE’ at top of this document to process and save data. This will take some time to run.

data <- processPgDetections(pps, mode='db', id='epacific_bw')
saveRDS(data, 'epacific_study.rds')
# Double check warning messages
print(getWarnings(data)$message)

If you have already run the processing code, ensure you have set ‘freshRun = FALSE’ at top of this document to read in the existing .rds file for downstream processing.

3. Assign species identity according to original PAMguard labels, then relabel for consistency across projects.

data <- setSpecies(data, 'pamguard')
reSpecies <- readRDS('species.rds')
print(reSpecies)
data <- setSpecies(data, 'reassign', value=reSpecies)

4. Filter out unwanted data: (1) subset and rename species for BANTER model and (2) filter to retain only Channel 1.

data <- setSpecies(data, method='reassign',
                   value=data.frame(old=c('ZC', 'MS', 'BB', 'BWC', 'BW37V'), new=c("Cuviers", "Stejnegers", "Bairds", "CrossSeamount", "Hubbs")))
goodSpecies<- c("Cuviers", "Stejnegers", "Bairds", "CrossSeamount", "Hubbs", "BW", "BW43", "possBW")
data <- filter(data, species %in% goodSpecies)
data_ch1only <- filter(data, Channel == '1')

5. Calculate Inter-Click Interval (ICI).

data_ch1only <- calculateICI(data_ch1only, time='peakTime')

6. Add GPS data: Add GPS data from PAMGuard table (2 hour threshold), then filter out data without GPS

data_ch1only <- addGps(data_ch1only, thresh = 7200)#11 events over the threshold
getWarnings(data_ch1only)
  
data_ch1only <- filter(data_ch1only, !is.na(Latitude))#filter out events w/o GPS
clicks <- getClickData(data_ch1only)
sum(is.na(clicks$Latitude))
unique(clicks[is.na(clicks$Latitude), 'eventId'])#confirm there are no remaining events w/o GPS

7. Add Environmental Data

#SST
data_ch1only <- matchEnvData(data_ch1only, nc='jplMURSST41mday', var='sst')
#Seafloor Depth
data_ch1only <- matchEnvData(data_ch1only, nc='erdSrtm30plusSeafloorGradient', var='sea_floor_depth')
#Seafloor Gradient
data_ch1only <- matchEnvData(data_ch1only, nc='erdSrtm30plusSeafloorGradient', var='magnitude_gradient')

7. Export Event Wav Files Wav files for events will be exported to a local drive, and then uploaded to Figshare.

#This should only be run one time, with access to original wav files. Will be commented out to ensure it is not accidentally run
#data <- readRDS(here('epacific', 'epacific_study.rds'))
#Check that all events are <120s
#checkStudy(data, maxLength = 120)

#Identify location of all recording files and log files
#data<- addRecordings (data, log = TRUE )
#saveRDS(data, 'epacific_study.rds')

#Identify directory to save wav files, and write event Wav Clips
#wavDir <- here('epacific', 'ccesEventWav')#identify directory to write clips to
#writeEventClips(data, buffer = 5, mode = 'event', useSample = FALSE, outDir = wavDir)
#saveRDS(data, 'epacific_study.rds')

8. Export to Annomate Export annotation information to a csv for upload to Ann-O-Mate. NOTE DO NOT SAVE THIS CODE CHUNK PUBLICLY–

#First, run:
unique(species(data)) #to see unique species and order in which they are named
#Then update line 119 to include full species names in order in which they occur
specMap <- data.frame(
    old = unique(species(data)),
    new = NA)
specMap$new <- c('Possible Beaked Whale', 'Ziphius cavirostris', 'Mesoplodon stejnegeri', 
                 'BW39V', 'Unidentified Beaked Whale', 'Berardius bairdii', 'BW43')
anno <- prepAnnotation(data, specMap = specMap, mode='event',
                       # following args are optional, fill in
                       # any that will be the same for all entries
                       source = 'figshare.com',
                       contact = 'shannon.rankin@noaa.gov')
#Export Annomate file, does this info save to acoustic study?

# to get figshare data you will need the article id and your personal token
# i recommend storing your token in an external file instead of typing it in here
# also DO NOT commit this file to any repository
figToken <- yaml::read_yaml('secrets.yaml')$fig_token
figId <- 13393360
figshareData <- getFigshareInfo(figToken, figId)

anno <- matchRecordingUrl(anno, figshareData)
# will give you warnings and messages about missing fields
checkAnnotation(anno)
# if you need to fix any manually, write to csv then read back in
annoFile <- 'AnnomateExport.csv'
write.csv(anno, file=annoFile, row.names = FALSE)
# go fix stuff
anno <- read.csv(annoFile, stringsAsFactors = FALSE)
# recheck to see if all seems fine
checkAnnotation(anno)
# add to acoustic study for storage
data <- addAnnotation(data, anno)
# this creates CSV ready for figshare, will also repeat messages from the check
export_annomate(data, file=annoFile)

9. Export data for BANTER (and drop species codes that will not be used for training). We will create two datasets: one with ICI and one without ICI, and save these for import into

banter_data <- export_banter(data_ch1only, dropSpecies = c('BW', 'possBW'), 
                         dropVars = c('All_ici', 'Click_Detector_3_ici', 'sea_floor_depth_mean', 'sst_mean', 'productivity_mean', 'chlorophyll_mean',  'magnitude_gradient_mean'), training=TRUE)
saveRDS(banter_data, file='epacific_banter_data.rds')

banter_data_ici <- export_banter(data_ch1only, dropSpecies = c("BW", "possBW"), dropVars = c('sea_floor_depth_mean', 'sst_mean', 'productivity_mean', 'chlorophyll_mean',  'magnitude_gradient_mean'),training=TRUE)
saveRDS(banter_data_ici, file='epacific_banter_data_ici.rds')

banter_data_env <- export_banter(data_ch1only, dropSpecies = c("BW", "possBW"), training=TRUE)
saveRDS(banter_data_env, file='epacific_banter_data_env.rds')

#save update of Acoustic Study
saveRDS(data_ch1only, 'epacific_study.rds')
saveRDS(data, 'epacificALL_study.rds')

Build a BANTER Classification Model

EC (only) Model

Initialize, Run & Evaluate Detector Model (stage 1).

banter_model_ec <- initBanterModel(banter_data$events)
banter_model_ec <- addBanterDetector(banter_model_ec, banter_data$detectors, ntree=1e4, sampsize=3, importance = TRUE)

plotDetectorTrace(banter_model_ec, detector = paste0('Click_Detector_', 1:3))
plotDetectorTrace(banter_model_ec, detector = paste0('Click_Detector_', 4:6))
summary(banter_model_ec)

Run BANTER Event Model (stage 2)

banter_model_ec <- runBanterModel(banter_model_ec, ntree=1e4, sampsize=4)
summary(banter_model_ec)

Once a stable model is identified, save model with tree/sampsize info in the filename.

saveRDS(banter_model_ec, 'epacific_banter_model_ec_t1e4s3_t1e4s4.rds')

ICI Model

Initialize, Run & Evaluate Detector Model (stage 1).

banter_model_ici <- initBanterModel(banter_data_ici$events)
banter_model_ici <- addBanterDetector(banter_model_ici, banter_data_ici$detectors, ntree=1e4, sampsize=3, importance = TRUE)

plotDetectorTrace(banter_model_ici, detector = paste0('Click_Detector_', 1:3))
plotDetectorTrace(banter_model_ici, detector = paste0('Click_Detector_', 4:6))
summary(banter_model_ici)

Run Event Model (stage 2)

banter_model_ici <- runBanterModel(banter_model_ici, ntree=1e4, sampsize=4)
summary(banter_model_ici)

Once a stable model is identified, save model with tree/sampsize info in the filename.

saveRDS(banter_model_ici, 'epacific_banter_model_ici_t1e4s3_t1e4s4.rds')

ENV Model

Initialize, Run & Evaluate Detector Model (stage 1).

banter_model_env <- initBanterModel(banter_data_env$events)
banter_model_env <- addBanterDetector(banter_model_env, banter_data_env$detectors, ntree=1e4, sampsize=1, importance = TRUE)

plotDetectorTrace(banter_model_env, detector = paste0('Click_Detector_', 1:3))
plotDetectorTrace(banter_model_env, detector = paste0('Click_Detector_', 4:6))
summary(banter_model_env)

Run Event Model (stage 2)

banter_model_env <- runBanterModel(banter_model_env, ntree=1e4, sampsize=5)
summary(banter_model_env)

Once a stable model is identified, save model with tree/sampsize info in the filename.

saveRDS(banter_model_env, 'epacific_banter_model_env_t1e4s1_t1e4s5.rds')

BANTER Analytics

There are a number of visualizations/data products that allow us to visualize our BANTER classifier; most use the rfPermute package (see BANTER Guidelines for more information.

First, load the models (if not a fresh run)

Identify the model you would like to examine (comment out the model you do not want to examine).

model_ec <- banter_model_ec
modelname_ec <- "banter_model_ec"

model_ici <- banter_model_ici
modelname_ici <- "banter_model_ici"

model_env <- banter_model_env
modelname_env <- "banter_model_env"

Extract the Random Forest model object from our BANTER model for analysis.

banter_model_ec_RF <- getBanterModel(model_ec)
banter_model_ici_RF <- getBanterModel(model_ici)
banter_model_env_RF <- getBanterModel(model_env)

Class Priors (Expected Error Rate)

epacific_ec_priors <- classPriors(banter_model_ec_RF, NULL)[,1]
epacific_ici_priors <- classPriors(banter_model_ici_RF, NULL)[,1]
epacific_env_priors <- classPriors(banter_model_env_RF, NULL)[,1]

Confusion Matrix

epacific_ec_confuseMatrix <- rfPermute::confusionMatrix(banter_model_ec_RF)
epacific_ec_confuseMatrix <- cbind(epacific_ec_confuseMatrix, priors = epacific_ec_priors)
epacific_ec_confuseMatrix <- kable(epacific_ec_confuseMatrix, align = "c", digits = c(0,0,0,0,2,2,2))%>%
  kable_classic()%>%
   column_spec(7, border_right = TRUE)%>%
   row_spec(0, bold = TRUE)%>%
   row_spec(4,hline_after = TRUE)%>%
   row_spec(7, bold = TRUE)%>%
   save_kable("../manuscript/manuscript_files/epacific_ec_confuseMatrix.png", zoom = 9)

epacific_ici_confuseMatrix <- rfPermute::confusionMatrix(banter_model_ici_RF)
epacific_ici_confuseMatrix <- cbind(epacific_ici_confuseMatrix, priors = epacific_ici_priors)
epacific_ici_confuseMatrix <- kable(epacific_ici_confuseMatrix, align = "c", digits = c(0,0,0,0,2,2,2))%>%
  kable_classic()%>%
  column_spec(7, border_right = TRUE)%>%
  row_spec(0, bold = TRUE)%>%
  row_spec(4,hline_after = TRUE)%>%
  row_spec(7, bold = TRUE)%>%
  save_kable("../manuscript/manuscript_files/epacific_ici_confuseMatrix.png", zoom = 9)

epacific_env_confuseMatrix <- rfPermute::confusionMatrix(banter_model_env_RF)
epacific_env_confuseMatrix <- cbind(epacific_env_confuseMatrix, priors = epacific_env_priors)
epacific_env_confuseMatrix <- kable(epacific_env_confuseMatrix, align = "c", digits = c(0,0,0,0,2,2,2))%>%
  kable_classic()%>%
  column_spec(7, border_right = TRUE)%>%
  row_spec(0, bold = TRUE)%>%
  row_spec(4,hline_after = TRUE)%>%
  row_spec(7, bold = TRUE)%>%
  save_kable("../manuscript/manuscript_files/epacific_env_confuseMatrix.png", zoom = 9)

BANTER Model E.Pacific EC Confusion Matrix

BANTER Model E.Pacific ICI Confusion Matrix

BANTER Model E.Pacific ENV Confusion Matrix

Proximity Plot

png(('../manuscript/manuscript_files/epacific_ec_proximity.png'), width = 20, height = 20, units = 'cm',  res = 300)
plotProximity(banter_model_ec_RF)
dev.off()
ec_epacific_proximityPlot <- plotProximity(banter_model_ec_RF)

png(('../manuscript/manuscript_files/epacific_ici_proximity.png'), width = 20, height = 20, units = 'cm',  res = 300)
ici_epacific_proximityPlot <- plotProximity(banter_model_ici_RF)

## Warning in MASS::cov.trob(data[, vars]): Probable convergence failure

dev.off()
ici_epacific_proximityPlot <- plotProximity(banter_model_ici_RF)

## Warning in MASS::cov.trob(data[, vars]): Probable convergence failure

png(('../manuscript/manuscript_files/epacific_env_proximity.png'), width = 20, height = 20, units = 'cm',  res = 300)
env_epacific_proximityPlot <- plotProximity(banter_model_env_RF)
dev.off()
env_epacific_proximityPlot <- plotProximity(banter_model_env_RF)

Importance Heatmap

png(('../manuscript/manuscript_files/epacific_ec_importance.png'), width = 30, height = 25, units = 'cm',  res = 300)
plotImportance(banter_model_ec_RF, plot.type="heatmap")
dev.off()
ec_epacific_importance <- plotImportance(banter_model_ec_RF, plot.type="heatmap")

png(('../manuscript/manuscript_files/epacific_ici_importance.png'), width = 30, height = 25, units = 'cm',  res = 300)
plotImportance(banter_model_ici_RF, plot.type="heatmap")
dev.off()
ici_epacific_importance <- plotImportance(banter_model_ici_RF, plot.type="heatmap")

png(('../manuscript/manuscript_files/epacific_env_importance.png'), width = 30, height = 25, units = 'cm',  res = 300)
plotImportance(banter_model_env_RF, plot.type="heatmap")
dev.off()
env_epacific_importance <- plotImportance(banter_model_env_RF, plot.type="heatmap")

PlotVotes

png(('../manuscript/manuscript_files/epacific_ec_votes.png'), width = 20, height = 20, units = 'cm',  res = 300)
plotVotes(banter_model_ec_RF)
dev.off()
epacific_votes <- plotVotes(banter_model_ec_RF)

png(('../manuscript/manuscript_files/epacific_ici_votes.png'), width = 20, height = 20, units = 'cm',  res = 300)
plotVotes(banter_model_ici_RF)
dev.off()
ici_epacific_votes <- plotVotes(banter_model_ici_RF)

png(('../manuscript/manuscript_files/epacific_env_votes.png'), width = 20, height = 20, units = 'cm',  res = 300)
plotVotes(banter_model_env_RF)
dev.off()
env_epacific_votes <- plotVotes(banter_model_env_RF)

Plot Predicted Probabilities

plotPredictedProbs(banter_model_ec_RF, bins=30, plot=TRUE)

plotPredictedProbs(banter_model_ici_RF, bins=30, plot=TRUE)

plotPredictedProbs(banter_model_env_RF, bins=30, plot=TRUE)

Create Figure for Publication

confuse <- magick::image_read(here('manuscript', 'manuscript_files', 'epacific_ec_confuseMatrix.png'))%>%
  image_border(color="#ffffff", geometry = "50x130")%>%
  image_annotate("a) Confusion Matrix", size=100, color = "black")
vote <- magick::image_read(here('manuscript', 'manuscript_files', 'epacific_ec_votes.png'))%>%
  image_border(color="#ffffff", geometry = "270x130")%>%
  image_annotate("d) Vote Plot", size=100, color = "black")
prox <- magick::image_read(here('manuscript', 'manuscript_files', 'epacific_ec_proximity.png'))%>%
  image_border(color="#ffffff", geometry = "270x130")%>%
  image_annotate("b) Proximity Plot", size=100, color = "black")
heat <- magick::image_read(here('manuscript', 'manuscript_files', 'epacific_ec_importance.png'))%>%
  image_border(color="#ffffff", geometry = "270x130")%>%
  image_scale("3300")%>%
  image_annotate("d) Importance Heat Map", size=100, color = "black")
  
epacific_ec_Figure <-image_append(c(prox, heat, vote))
epacific_ec_Figure<- image_append(c(confuse, epacific_ec_Figure), stack=TRUE)
image_write(epacific_ec_Figure, path = here('manuscript', 'manuscript_files','epacific_ec_Figure.png'), format ='png')
print(epacific_ec_Figure, info=FALSE)

confuse <- magick::image_read(here('manuscript', 'manuscript_files', 'epacific_ici_confuseMatrix.png'))%>%
  image_border(color="#ffffff", geometry = "50x130")%>%
  image_annotate("a) Confusion Matrix", size=100, color = "black")
vote <- magick::image_read(here('manuscript', 'manuscript_files', 'epacific_ici_votes.png'))%>%
  image_border(color="#ffffff", geometry = "270x130")%>%
  image_annotate("d) Vote Plot", size=100, color = "black")
prox <- magick::image_read(here('manuscript', 'manuscript_files', 'epacific_ici_proximity.png'))%>%
  image_border(color="#ffffff", geometry = "270x130")%>%
  image_annotate("b) Proximity Plot", size=100, color = "black")
heat <- magick::image_read(here('manuscript', 'manuscript_files', 'epacific_ici_importance.png'))%>%
  image_border(color="#ffffff", geometry = "270x130")%>%
  image_scale("3300")%>%
  image_annotate("d) Importance Heat Map", size=100, color = "black")
  
epacific_ici_Figure <-image_append(c(prox, heat, vote))
epacific_ici_Figure<- image_append(c(confuse, epacific_ici_Figure), stack=TRUE)
image_write(epacific_ici_Figure, path = here('manuscript', 'manuscript_files','epacific_ici_Figure.png'), format ='png')
print(epacific_ici_Figure, info=FALSE)

confuse <- magick::image_read(here('manuscript', 'manuscript_files', 'epacific_env_confuseMatrix.png'))%>%
  image_border(color="#ffffff", geometry = "50x130")%>%
  image_annotate("a) Confusion Matrix", size=100, color = "black")
vote <- magick::image_read(here('manuscript', 'manuscript_files', 'epacific_env_votes.png'))%>%
  image_border(color="#ffffff", geometry = "270x130")%>%
  image_annotate("d) Vote Plot", size=100, color = "black")
prox <- magick::image_read(here('manuscript', 'manuscript_files', 'epacific_env_proximity.png'))%>%
  image_border(color="#ffffff", geometry = "270x130")%>%
  image_annotate("b) Proximity Plot", size=100, color = "black")
heat <- magick::image_read(here('manuscript', 'manuscript_files', 'epacific_env_importance.png'))%>%
  image_border(color="#ffffff", geometry = "270x130")%>%
  image_scale("3300")%>%
  image_annotate("d) Importance Heat Map", size=100, color = "black")
  
epacific_env_Figure <-image_append(c(prox, heat, vote))
epacific_env_Figure<- image_append(c(confuse, epacific_env_Figure), stack=TRUE)
image_write(epacific_env_Figure, path = here('manuscript', 'manuscript_files','epacific_env_Figure.png'), format ='png')
print(epacific_env_Figure, info=FALSE)